The genus Bordetella comprises four species Bordetella pertussis, Bordetella parapertussis, Bordetella bronchiseptica and Bordetella avium. 
The bordetellae are Gram-negative coccobacilli responsible for respiratory infections. Bordetella pertussis and Bordetella parapertussis, agents of whooping cough, are strictly human pathogens. Bordetella bronchiseptica is pathogenic for various mammals, and more rarely for man, and, in distinction to B. pertussis and B. parapertussis, is capable of surviving outside the host. Bordetella avium is pathogenic only for birds.
Since the introduction of whooping cough vaccination in countries where vaccine cover is greater than 80%, it has been possible to observe a dramatic fall in morbidity and mortality. This fall is indeed attributable to vaccination since, in several countries (Great Britain, Sweden, Japan, etc.), deadly epidemics of whooping cough have taken place in the years following the cessation of vaccination.
The invention affords immunogenic compositions which can participate in the making of whooping cough vaccines, these compositions being at least partly of the xe2x80x9cacellularxe2x80x9d type and displaying an efficacy at least identical to that of the known vaccine.
The invention relates, on the one hand to vaccines which can be used in veterinary medicine, and on the other hand to vaccines which can be used in human medicine.
The whooping cough vaccine in current use is a cellular vaccine composed of heat-inactivated bacterial suspensions of B. pertussis (mixture of two strains differing in the expression of agglutinogens). This vaccine is generally used in combined form with purified diphtheria and tetanus fractions, the hemophilus and the inactivated polio viral component. Vaccination consists of three injections at one-month intervals from the age of two months and an injection at 18 months. No other booster injection is performed thereafter.
This vaccine is sometimes poorly tolerated, both locally and generally. It has, in particular, been blamed for giving rise to serious neurological complications of the acute encephalitis type; however, very recent studies appear to conclude that there is no statistical proof of a relationship between the cellular vaccine and the severe neurological complications (Griffiths A H. Vaccine 1989; 7:199-210).
It remains nonetheless true that the cellular vaccine is poorly tolerated and is responsible for reversible, but undesirable, effects. For these reasons, a new vaccine lacking these effects is desirable. In order to envisage the defining of a new vaccine, it appeared necessary to characterize certain factors involved in the virulence of the bacterium and, where appropriate, in the regulation of the virulence. When purified, each and every one of these different factors is a theoretical candidate for the making of a whooping cough vaccine termed xe2x80x9cacellularxe2x80x9d as opposed to the traditional vaccine. This new type of vaccine should provide, in addition to better tolerance, an efficacy at least equal to that of the traditional vaccine.
The factors involved in the virulence of a B. pertussis have been identified as follows: whooping cough may be defined in broad outline by the association of an infectious syndrome, involving the adhesion of the bacteria to the target cells (ciliated cells of the respiratory apparatus), without invasion or dissemination in the host""s body, and of a secondary toxin-induced syndrome including the local cytopathogenic effects which are elective for the ciliated respiratory epithelium (destruction and removal of ciliated cells, accumulation of mucus, inflammatory reaction) and systemic effects, the most obvious of which is hyperleucocytosis with hyperlymphocytosis.
As a result of recent techniques in molecular biology, a number of factors involved in the virulence of B. pertussis have been characterized and the regulation of their expression understood. These factors may be classified in two categories, those participating in the infectious syndrome (adhesins) and those playing a part in the toxin-induced syndrome (toxins).
The Adhesins
filamentous hemagglutinin or FHA is considered to play a major part in the adhesion of the bacterium to the ciliated epithelium (Locht C., Bertin P., Menozzi F. D. and Renaud G. Mol. Microbiol. 1993, 9:653-66). FHA is always expressed by the virulent strains and is secreted. Its structural gene has been cloned and sequenced (Relman D. et al., 1989, Proc. Natl. Acad. Sci. USA, 86:2637-2641). It codes for a protein of 360 kDa, but only a 220-kDa fragment can be purified. This protein binds to the glycoproteins of ciliated cells and possesses binding sites for the integrins of lymphocytes and macrophages. It has just been shown recently that FHA displays a homology with certain proteins of the host""s endothelial cells (Tuomanen E., Prasad S. M., George J. S., Hoepelman A. I. M., Ibsen P., Heron I., and Starzyk R. M. 1993. Proc. Natl. Sci. USA. 90:7824-7828).
The two agglutinogens or AGGs of B. pertussis enable strains to be classified in serotypes. Two AGGs have been characterized. These proteins are secreted and play a part in the adhesion of the bacterium to the epithelial cells (Mooi F. Van der Heide H. G. D., Ter Avest A. R., Welinder K. G., Livey I., Van der Zeijst B. A. M., and Gaastra, W. 1987. Microb. Pathog. 2:473-484).
Pertactin or PRN is a protein of 93 kDa, but only a 69-kDa fragment can be purified. This protein possesses two binding sites for the integrins of macrophages and of lymphocytes (Charles I, Dougan G., Pickard D., Chattfield S. Smith M. Novotny P., Morissey P. and Fairweather N. F. 1989 Proc Natl Acad Sci. 86: 3554-3558).
Pertussis toxin or PTX, a secreted type A-B toxin which, besides its cytopathogenic effects, participates in adhesion via its B subunit. The B oligomer is capable of binding to the receptors of ciliated cells, but not necessarily to the same receptors as those for FHA. The binding of PTX to leukocytes would appear to prevent their migration to the site of the inflammatory reaction. This binding appears to induce an increase in the number of functional integrin molecules to leukocytes, thereby promoting binding of the bacterium via FHA (Rozindski E., Burnette W. N., Jones T., Mar V., and Tuomanen E. 1993 J. Exp. Med. 178:917-924).
The Toxins
Pertussis toxin or PTX is secreted and considered to be the major toxin of B. pertussis. Its A subunit possesses ADP-ribosyl transferase activity. After binding of the B portion of the toxin to the target cell, this A subunit is capable of entering the cell, of inactivating the regulatory G proteins and thus of causing interference with all cellular functions. It is this factor which appears to be responsible for the generalized biological effects observed during the disease, such as hyperlymphocytosis, hyperinsulinemia and sensitivity to histamine.
Dermonecrotic toxin or DNT, which has not yet been well characterized, and tracheal cytotoxin or TCT, a secreted small glycoprotein of the muramyl peptide family, derived from the peptidoglycan of the bacterium, appear to act in concert to destroy the ciliated cells of the host""s respiratory apparatus. TCT prevents, in addition, the regeneration of the respiratory epithelium (Luker K., Collier J. L., Kolodziej E. W., Marshall G. R., and Goldman W. E. 1993. Proc. Natl. Acad. Sci. USA. 90:2365-2369).
Adenyl cyclase-hemolysin or Ac-Hly is a bifunctional protein possessing adenyl cyclase activity and hemolytic activity. It is secreted by the bacterium. Its structural gene has been cloned and sequenced (Glaser P. et al., 1988, Molec. Microb. 2, 19-20). This protein has been found to belong to the family of toxins termed xe2x80x9cRTXxe2x80x9d for xe2x80x9crepeats in toxinsxe2x80x9d, and displays homologies with the hemolysin of Escherichia coli and of Actinobacillus pleuropneumoniae and the leukotoxins of Pasteurella haemolytica and of Actinobacillus actinomycetemcomitans. This protein, like PTX, is capable of entering eukaryotic cells such as macrophages, of being activated by calmodulin, of synthesizing large amounts of CAMP and of interfering with the cellular functions (Coote J. 1992. FEMS Microbiol. Rev. 88:137-162).
Similarly, the factors involved in the virulence of B. parapertussis and B. bronchiseptica have been identified.
B. pertussis, B. parapertussis and B. bronchiseptica infections are indistinguishable from a clinical standpoint. These bacteria have more than 75% homology in respect of the DNA. They have been classified in species only on the basis of phenotypic and biochemical characters. B. parapertussis and B. bronchiseptica synthesize virulence factors functionally and immunologically very close to B. pertussis, with the exception of PTX.
A vaccine composed of bacterial suspensions of inactivated B. pertussis protects against a B. pertussis infection, but also against a B. parapertussis and a B. bronchiseptica infection in the mouse model. Although there are no epidemiological data on B. parapertussis infections in France, it should be noted that few strains of this species are isolated in this country, a country which has been vaccinated for 25 years with a xe2x80x9cpertussis vaccinexe2x80x9d, whereas they are isolated in unvaccinated or poorly vaccinated countries.
Besides the presence of these different adhesins and toxins, the bordetellae are characterized by a regulation of the expression of the factors involved in their virulence. In other words, the bordetellae undergo phase variations and modulations.
The bordetellae, depending on their environment, may become xe2x80x9cavirulentxe2x80x9d, that is to say incapable of inducing lethality, an inflammatory reaction and pulmonary lesions in the mouse model of respiratory infection. They undergo either a phase modulation or a phase variation. The phase variation is observed at a frequency ranging from 10xe2x88x923 to 10xe2x88x926 and is almost irreversible. It manifests itself in a cessation of the expression of the toxins and adhesins described above and in the expression of other factors which have not yet been well characterized (change of Phase I xe2x80x9cvirulentxe2x80x9d bacteria to Phase IV xe2x80x9cavirulentxe2x80x9d bacteria). The Phase I and Phase IV bacteria have been described by Lacey B. 1960, J. Hyg. 58:57-93. The phase modulation, phenotypically similar to the phase variation, is completely reversible and manifests itself in a cessation of the synthesis of the adhesins and the toxins when there are environmental changes (composition of the culture medium, temperature, etc.).
The phase variation and phase modulation observed in Bordetella are under the control of two regulatory genes bvg A and bvg S (Arico B. et al., 1989, Proc. Natl. Acad. Sci USA, 86: 6671-6675).
The bvg S gene codes for a protein sensitive to external conditions. This protein modulates by phosphorylation the activity of the protein encoded by the bvgA gene, which is, on the one hand a positive activator of the transcription of the genes coding for the virulence factors (vag genes for xe2x80x9cvir activated genesxe2x80x9d) mentioned above (Uhl M. A. and Miller J. 1994. Proc. Natl. Acad. Sci USA 91:1163-1167), and on the other hand a repressor of the transcription of certain genes (Beattie D. T. et al., J. of Bacteriology, January 93, p. 159-527). The genes whose expression is repressed are referred to as vrg genes for xe2x80x9cvir repressed genesxe2x80x9d, and are still poorly characterized. It has, however, been shown that the vrg 6 gene of B. pertussis codes for a protein having a role in the peristance of the bacterium in the host (Beatties D. et al., 1992, Infect. Imm. 60:571-577). In B. bronchiseptica, two proteins encoded by the vrg genes have been characterized: they are proteins of the flagella type (Phase I B. bronchiseptica is an immobile bacterium which does not synthesize flagella but synthesizes adhesins and toxins, and Phase IV B. bronchiseptica is a mobile bacterium which synthesizes flagella).
In order to measure the virulence of the bacteria and to evaluate the local and general toxinic effects, a mouse model of respiratory infection has been developed (Guiso N. et al., 1991, Microb. Pathogen 11, 423-431). Using this mouse model, it has been possible to show that chemically or genetically inactivated PTX is a good immunogen. This anatoxin has a protective activity against lethal B. pertussis infections, but does not appear to induce the synthesis of effective antibodies against the persistance of the bacterium (Khelef. N. Danve B. Quentin-Millet M. J. and Guiso N. 1993 Infect Immun. 64:486-490).
These results relating to the virulence of bordetellae and to the regulation of this virulence show that whooping cough is a multifactorial disease and that the vaccine must not only protect against lethal infections but also against the persistance of the bacterium. Similar conclusions apply to the infections due to B. parapertussis or to B. bronchiseptica. 
Attempts to develop an acellular vaccine from the isolated components of the adhesin or toxin family have been performed. Thus, acellular compositions containing either purified B. pertussis toxin (PTX) or this toxin combined with purified filamentous hemagglutinin (FHA) have been prepared.
The first trials of tolerance of these acellular compositions (PTX or PTX-FHA) in man show a marked decrease in both local complications (pain, swelling) and general complications (fever, convulsions, etc.) in comparison with the traditional cellular vaccine (Edward K., J. Infect. Dis. 1993, 168, 15-20).
These new preparations (PTX or PTX-FHA) have good immunogenicity and induce a high level of antibodies. However, the test for vaccinal antibodies is an imperfect method, since seroconversion is not synonymous with protection against the disease, and no demonstration has been made of the protective character of the antibodies obtained, or of the possible level of protection.
The results of clinical trials of different cellular and acellular vaccines have been published (International Symposium on Pertussis Vaccine trials, Rome 30.10.95-1.11.95). These results show that not all the cellular vaccines are equivalent, some are very effective and induce few side effects and others are of very low efficacy and induce greater side effects.
The published results show that the acellular vaccines tested, monovalent (PTX), bivalent (PTX, FHA), trivalent (PTX, FHA, PRN) or pentavalent (PTX, FHA, PRN, AGG2, AGG3) induce very few side effects, are all immunogenic and all have an efficacy against the disease (according to WHO definition) which is greater than or equal to 70%. However, the efficacy of an acellular PTX-FHA vaccine is always less, irrespective of the particular definition which may be used, than that of an effective cellular vaccine.
In spite of the encouraging results obtained as regards the immunogenicity of different compositions containing both adhesins and toxins of B. pertussis, the inventors considered that an effective protection against the disease due to B. pertussis, a parapertussis or B. bronchiseptica infection necessitated consideration of additional factors with respect to the adhesins and toxins, and especially factors participating in the persistance of the bacterium.
Their observations led them to define a model described in detail in the experimental part, from which new criteria prior to the definition of vaccines have been defined.
Thus, according to the present application, to define effective and substantially nontoxic vaccines against at least one of the bordetellae B. pertussis, B. parapertussis or B bronchiseptica, it is appropriate to employ not only one or more adhesins and/or toxins of these bacteria, but also one or more factors whose synthesis is repressed when there is expression of the toxins and adhesins of the bacterium. These factors are, in particular, expression products of the vrg genes which have been referred to above.
With the object of preparing a whooping cough vaccine lacking side effects and which is protective against the local or systemic effects due to the toxins synthesized by B. pertussis and/or B. parapertussis and/or B bronchiseptica, and protective against the persistance of these bacteria in the host, the subject of the invention is an immunogenic composition, characterized in that it comprises an adenyl cyclase-hemolysin (AC-Hly) protein, or an immunogenic portion of this AC-Hly, characteristic of a strain of Bordetella chosen from B. pertussis, B. parapertussis or B bronchiseptica, and in that it comprises, in addition, a bacterial extract containing the products of the vrg genes of a strain of Bordetella chosen from B. pertussis, B. parapertussis or B bronchiseptica, or a portion of these expression products which is sufficient to induce an immune response in a host to which the extract might be administered.
None of the acellular vaccines prepared to date contain AC-Hly. Now, according to the inventors, this protein should be incorporated in a vaccinating composition since it plays an important part in the virulence of the bordetellae for the following reasons:
it is expressed very early after infection, since specific antibodies are synthesized and detected, both in infected humans and in infected animals, from the beginning of the infection (Khelef N, Sakamoto H., and Guiso N 1992. Microbiol. Pathogen. 12:227-235 and Gueirad P. and Guiso N. 1993 Infect. Immun. 61:4072-4078).
it is necessary to the bacterium in order to initiate infection (Khelef N. Sakamoto H. and Guiso N. 1992 Microbiol. Pathogen. 12:227-235).
its use in purified form as antigen in the mouse model of respiratory infection protects the mice against colonization of the respiratory apparatus (Guiso, Szatanik and Rocancourt. 1989 Microb. Pathogen. 11:423-431).
it is responsible in vitro for the death of alveolar macrophages by apoptosis (Khelef, Zyglinski and Guiso Infect. Immun. 61:4064-4071). The other adhesins or toxins do not participate in the apoptosis process.
Moreover, the presence in the immunogenic composition of a bacterial extract comprising the expression products of the vrg genes would enable the humoral and/or cellular immune response obtained after infection to be improved in vaccinated subjects and would also contribute to the protection against the persistance of the bacterium.
The bacterial extract termed xe2x80x9cvrg bacterial extractxe2x80x9d, which has been referred to above, contains all the constituents of the outer membrane, including the LPS endotoxin, of a phase IV bacterium, that is to say of a bacterium not expressing the vag genes. The LPS endotoxin may alternatively be removed or detoxified.
This extract may be present in suspension form.
A first preferred immunogenic composition of the invention is an immunogenic composition which is characterized in that it comprises, in addition, one or more adhesins or toxins, respectively, of B. pertussis, B. parapertussis or B bronchiseptica, chosen from FHA, the AGGs or PRN and PTX.
The adhesins of the bordetellae strengthen the immunogenic character of the composition containing AC-Hly and the vrg bacterial extract.
According to a first embodiment of the invention, the immunogenic composition is characterized in that it comprises the PTX toxin of B. pertussis, the AC-Hly toxin of B. pertussis, and the FHA and PRN adhesins, and a bacterial extract containing the proteins encoded by the vrg genes of B. pertussis, or a portion of these proteins which is sufficient to induce an immune response in a host to which the extract might be administered.
Such a composition may be employed for preparing vaccines which are protective against the lethal and systemic effects of B. pertussis in man and against the persistance of thp bacterium.
According to a preferred embodiment of the invention, the AC-Hly toxin of B. pertussis is also contained in a bacterial extract containing all or part of the virulence factors of the adhesin or toxin group of B. pertussis. 
The xe2x80x9cvagxe2x80x9d or xe2x80x9cvrgxe2x80x9d bacterial extracts which are employed for implementing the invention are preferably extracts termed xe2x80x9curea extractsxe2x80x9d.
A xe2x80x9curea extractxe2x80x9d is composed of a mixture of proteins expressed at the surface of the bacterium and which are separated from the bacterium after incubation of the latter with 5M urea. The xe2x80x9cvag urea extractxe2x80x9d of B bronchiseptica, for example, contains, among other proteins, AC-Hly, FHa, PRN and LPS (lipopolysaccharide endotoxin), and the xe2x80x9cvrg urea extractxe2x80x9d contains several proteins not yet characterized, the flagella and LPS.
The use of urea extracts makes it possible, in particular, to produce a vaccine which is cheaper compared to a vaccine which would be obtained from the proteins contained in the extracts, in purified form.
In addition, the inventors found that the urea extracts used can induce a T type cellular immune response (lymphoproliferation), thus behaving like the cellular vaccine used hitherto.
On the contrary, exclusively acellular compositions would not induce a T response, which reaction, however, occurs in the case of infection.
The vag or vrg urea extracts are prepared, respectively, from phase I or phase IV bacteria. Where appropriate, the phase IV bacteria are replaced by bacteria whose bvgS gene is mutated in such a way that the bacteria express only the proteins encoded by the vrg genes. In embodiments, the bacterial extract expressing the virulence repression proteins encoded by the vrg genes of a strain of Bordetella chosen from B. pertussis, B. parapertussis, or B. bronchiseptica, or a portion of these proteins which is sufficient to induce an immune response in a host to which the extract might be administered, is replaced, respectively by a B. pertussis, B. parapertussis, or B. bronchiseptica strain whose bvgS gene is mutated in such a way that the bacterium essentially expresses the vrg genes to the detriment of the vag genes.
The preparation of these extracts is described in detail in the experimental part.
Thus the invention relates preferentially to an immunogenic composition comprising both a vag urea extract of B. pertussis and a vrg urea extract of B. pertussis. 
A B. pertussis strain which is suitable for the preparation of these extracts is the strain HAV falling within the scope of the invention and deposited at the CNCM (Collection Nationale de Cultures de Microorganismes [National Collection of Microorganism Cultures] in Paris) on Oct. 19, 1994 under No. I-1485. To prepare the vag urea extract, the strain HAV may be used directly since it is a phase I strain.
In contrast, the vrg urea extract is obtained from a phase IV strain derived from the phase I strain, for example by mutation of the bvgS gene of the bacterium or by culturing said phase I strain in a medium containing only magnesium sulfate so as to obtain the expression of only the vrg genes of B. pertussis. 
The invention affords, in addition, immunogenic compositions prepared, according to the principles described above for B. pertussis, from the bordetellae of the B. parapertussis or B. bronchiseptica family.
It is, in effect, known that vaccination with the cellular vaccine available on the market, that is to say composed of bacterial suspensions of inactivated phase I B. pertussis bacteria, protects against B. pertussis and B. parapertussis infections in the mouse model. However, it has recently been demonstrated that administration of the purified B. pertussis factors (PTX or FHA or PRN) does not protect against the disease and the infection due to B. parapertussis. Similarly, the administration of the purified AC-Hly of B. parapertussis does not protect against a B. pertussis infection in the mouse model.
These results suggest that, in spite of a very high homology between the two species, the protection is species-specific. Hence the use of acellular vaccines composed of PTX, PRN and FRA of B. pertussis strains alone as whooping cough vaccines in years to come would incur the risk of causing an increase in B. parapertussis infections (Khelef Danve Quentin-Millet and Guiso 1993, Infect. Immun. 61:46-490 and Gueirad P., and Guiso 1993 Infect. Immun 61:4072-4078).
Hence the invention also affords immunogenic compositions which are characterized in that they comprise the AC-Hly toxin of B. parapertussis and a bacterial extract containing all or part of the proteins encoded by the vrg genes of B. parapertussis. 
Advantageously the AC-Hly toxin of B. parapertussis is contained in a bacterial extract containing all or part of the virulence factors of the adhesin or toxin group of B. parapertussis. Such an extract is preferably a vag urea extract obtained, for example, from B. parapertussis strain No. 1 deposited at the CNCM on Dec. 2, 1994 under No. I-1498. B. parapertussis strain No. I-1498 falls within the scope of the present invention.
Similarly the vrg bacterial extract is preferably a vrg urea extract obtained, for example, from B. parapertussis strain No. 1, deposited at the CNCM on Dec. 2, 1994 under No. I-1498, according to procedures similar to those which have been described for the B. pertussis strain.
According to another aspect, and especially with the object of producing a veterinary vaccine, the present application affords immunogenic compositions which are characterized in that they comprise the AC-Hly toxin of B. bronchiseptica and a bacterial extract containing all or part of the proteins encoded by the vrg genes of B. bronchiseptica. 
As mentioned above in connection with B. pertussis and B. parapertussis, the AC-Hly toxin of B. bronchiseptica is contained in a bacterial extract comprising all or part of the virulence factors (adhesins and/or toxins) of B. bronchiseptica. 
The extract in question is advantageously a vag urea extract obtained, for example, from B. bronchiseptica strain 973S deposited at the CNCM on May 12, 1989 under No. I-858.
Similarly, the vrg urea extract is preferably obtained from B. bronchiseptica strain 973S, deposited at the CNCM on May 12, 1989 under No. I-858, according to the procedures described above for B. pertussis. 
According to another particular embodiment of the invention, an immunogenic composition as described above which is capable of inducing the production of antibodies against B. bronchiseptica comprises polypeptide components characteristic of the flagella of B. bronchiseptica. These components have been described by Akerley B. J. et al. (J. of Bact. February 1992, p. 980-990).
Antibordetella humoral immunity has been considered for a long time to be the only important one. In effect, it has been known for a very long time that, after infection, anti-PTX and anti-FRA antibodies may be detected in the serum of patients. Furthermore, it is possible to protect passively against the disease, in the mouse model, with anti-PTX antibodies. The circulating antibodies hence play a part in the neutralization of PTX and in the inhibition of the attachment of the bacterium.
However, a correlation has never been demonstrated between the level of antibodies specific for these antigens in the serum of vaccinated or infected individuals and protection against the disease. Very little is known regarding the immunity of the other factors, especially AC-Hly, PRN and LPS. It has, however, been shown
that it is possible to protect mice passively against a B. pertussis infection with anti-AC-Hly antibodies.
that, in the serum of unvaccinated infected children (above 8 months of age in order to avoid the presence of maternal antibodies which may distort the interpretation of the serology, and below two years of age in order to know the clinical history) or sera of mice infected with clinical isolates of B. pertussis or of B bronchiseptica, it is possible to detect anti-AC-Hly antibodies, very early, indicating that this protein is expressed in vivo and from the beginning of the infection, anti-FHA antibodies, anti-PTX antibodies and anti-LPS antibodies. Anti-PRN antibodies appear much later and do not persist for very long (Guiso, Grimprel, Anjak and Bxc3xa9guxc3xa9 1993, Eur. J. Clin. Microbiol. Infect. Dis. 12:596-600).
Vaccination with the cellular vaccine also induces the synthesis of antibodies against these factors, whereas the acellular vaccines tested to date induce antibodies only against the factors of which they are composed.
Various studies suggest, in addition, that cellular immunity is also required for protection against infection. Several teams have shown that bordetellae can enter different cell types, epithelial cells, lymphocytes and monocytes, and can persist therein in vitro. Moreover, following epidemiological studies, it has been observed that B. bronchiseptica can persist for a very long time in man. Lastly, although it is not possible to isolate bacteria more than 40 days after infection in the lungs in mice, it is still possible to detect DNA by PCR. These results collectively suggest that bordetellae might persist intracellularly in the host at a given instant in the infection, and that cellular immunity might be required in order to control an intracellular state which would thus enable the bacteria to evade the host""s immune defences.
The demonstration of T cell involvement during infection has been obtained recently (Redhead, Watkins, Barnard and Mills 1993, Infect. Immun. 61:3190-3198). In effect, at a sublethal dose, normal mice eliminate the bacteria in 30 to 40 days, whereas T cell-deficient mice (xe2x80x9cnudexe2x80x9d mice) are incapable of eliminating the bacteria and develop a chronic infection. The transfer of T cells from convalescent mice to xe2x80x9cnudexe2x80x9d mice renders the latter capable of eliminating the bacteria. Furthermore, the spleen cells of convalescent mice produce high levels of IL-2, IFN-xcex3 and TNF, but not of IL-4 and IL-5. This profile is characteristic of the T-cell subpopulations which are the Th1 type cells. The immunization of mice with the whole vaccine also induces a Th1 type response and a moderate antibody response, whereas immunization with an acellular vaccine composed of PTX, FHA and PRN induces a Th2 type response and a very high level of antibodies. While there is a strong correlation between high levels of serum IgG specific for B. pertussis antigens and an elimination of the bacterium from the lungs, direct cellular responses are, however, necessary for a complete elimination of the bacterium. The slower elimination of the bacteria in mice immunized with acellular vaccines than with whole vaccines which has been observed hence appears to be due to the fact that the acellular vaccine does not promote the induction of Th1 cells. Irrespective of the mechanism of action of Th1, it is hence now clear that the T cells play an important part, not only indirectly in stimulating antibody synthesis, but also directly in anti-B. pertussis immunity via the recruitment, stimulation and activation of phagocytic cells such as macrophages and neutrophilic polymorphonuclear leukocytes.
It has, in addition, been observed that bordetellae can persist in man, especially B. bronchiseptica. In some cases, the bacterium persists for several weeks, in others several months. During some infections, the isolates have a different appearance. This different appearance corresponds to the cessation either of AC-Hly synthesis, or of the synthesis of all the factors encoded by the vag genes.
The inventors observed, most advantageously in the context of the search for an effective protection against the disease caused by an infection due to bordetellae and against the persistence of these bacteria in the host, that the bacterial extracts according to the invention cause a humoral immune response and a cellular immune response.
They showed that, after infection with B. bronchiseptica, there is induction of a humoral immunity and of a cellular immunity, as in the case of a B. pertussis infection. Furthermore, after vaccination with purified AC-Hly, there is induction of a humoral and cellular type immunity similar to that induced after infection or reinfection.
The subject of the invention is also vaccinating compositions comprising as active principle an immunogenic composition corresponding to one of the definitions given on the foregoing pages, in combination with a pharmaceutically acceptable vehicle and, where appropriate, with an adjuvant.
Like the whooping cough vaccines currently available on the market, the vaccine according to the invention may be combined with other vaccinating active principles, for example those of the vaccine against diphtheria, polio or diseases caused by Haemophilus or, generally speaking, with any immunogenic constituent, for example a particular inactivated pathogenic agent or toxin.
A vaccinating composition according to the invention can be species-specific and consequently capable of inducing protection against B. pertussis or B. parapertussis or B. bronchiseptica. On the contrary, it can be a mixture comprising as active principle an immunogenic composition against B. pertussis, as defined above, and an immunogenic composition against B. parapertussis. 
According to another embodiment of the invention, the vaccinating composition contains the immunogenic compositions prepared according to the invention against B. pertussis and against B. paravertussis and against B. bronchiseptica. 
An especially preferred vaccine against B. pertussis is characterized in that it comprises as active principle a xe2x80x9cvag urea extractxe2x80x9d of B. pertussis and a xe2x80x9cvrg urea extractxe2x80x9d of B. pertussis, the B. pertussis strain used to prepare these extracts preferably being the strain HAV deposited at the CNCM under No. I-1485.
An especially preferred vaccine against B. parapertussis is characterized in that it comprises as active principle a xe2x80x9cvag urea extractxe2x80x9d of B. parapertussis and a xe2x80x9cvrg urea extractxe2x80x9d of B. parapertussis, the B. parapertussis strain used to prepare these extracts preferably being the strain No. 1 deposited at the CNCM under No. I-1498.
An especially preferred vaccine against B. bronchiseptica is characterized in that it comprises as active principle a xe2x80x9cvag urea extractxe2x80x9d of B. bronchiseptica and a xe2x80x9cvrg urea extractxe2x80x9d of B bronchiseptica, the B. bronchiseptica strain used to prepare these extracts preferably being the strain 973S deposited at the CNCM under No. I-858.
Any vaccinating composition comprising a mixture of the vag urea extracts and the vrg urea extracts of B. pertussis and/or of B. parapertussis and/or of B. bronchiseptica also falls within the scope of the invention.
Where appropriate, the vag and vrg urea extracts employed are prepared from several isolates of the Bordetella species used.
The invention also affords a vaccine comprising as active principle B. pertussis strains expressing the vag genes and B. pertussis strains expressing the vrg genes, and/or B. parapertussis strains expressing the vag genes and B. parapertussis strains expressing the vrg genes, and/or B. bronchiseptica strains expressing the vag genes and B. bronchiseptica strains expressing the vrg genes.
The invention also relates to the bacterial extracts of the xe2x80x9curea extractsxe2x80x9d type as are obtained by carrying out the method described in detail in the experimental part.
Immunogenic compositions or vaccinating compositions in which AC-Hly is present in pure form, or in which it is replaced by a polypeptide comprising her C-terminal portion and/or a polypeptide comprising its internal portion, also fall within the scope of the invention.
The AC-Hly may be isolated from Bordetella or prepared, for example, by genetic engineering techniques.
The subject of the invention is also antibodies as are produced in a host to which immunogenic or vaccinating compositions as are described above, or a bacterial extract, have/has previously been administered.